Merge pull request #4329 from inception-project/bugfix/4328-HTML-file…

…s-are-not-rendered-if-they-use-the-HTML-namespace #4328 - HTML files are not rendered if they use the HTML namespace
inception-project · Nov 30, 2023 · 1d370df · 1d370df
2 parents 4b1b180 + 19b9fd8
commit 1d370df
Show file tree

Hide file tree

Showing 6 changed files with 52 additions and 7 deletions.
diff --git a/...in/java/de/tudarmstadt/ukp/inception/externaleditor/policy/DefaultHtmlDocumentPolicy.yaml b/...in/java/de/tudarmstadt/ukp/inception/externaleditor/policy/DefaultHtmlDocumentPolicy.yaml
@@ -19,6 +19,7 @@ version: 1.0
 case_sensitive: false
 default_attribute_action: DROP
 default_element_action: DROP
+default_namespace: http://www.w3.org/1999/xhtml
 debug: true
 policies:
   - action: PASS

diff --git a/...ava/de/tudarmstadt/ukp/inception/externaleditor/policy/DefaultHtmlDocumentPolicyTest.java b/...ava/de/tudarmstadt/ukp/inception/externaleditor/policy/DefaultHtmlDocumentPolicyTest.java
@@ -41,7 +41,7 @@ void thatOverrideFileIsPickedUp(@TempDir Path aTemp) throws Exception
 
         var sut = new DefaultHtmlDocumentPolicy();
 
-        assertThat(sut.getPolicy().getElementPolicies()).hasSize(74);
+        assertThat(sut.getPolicy().getElementPolicies()).hasSize(148);
 
         write(policyFile.toFile(), "policies: []", UTF_8);
         assertThat(policyFile).exists();
@@ -54,6 +54,6 @@ void thatOverrideFileIsPickedUp(@TempDir Path aTemp) throws Exception
 
         Files.delete(policyFile);
         assertThat(policyFile).doesNotExist();
-        assertThat(sut.getPolicy().getElementPolicies()).hasSize(74);
+        assertThat(sut.getPolicy().getElementPolicies()).hasSize(148);
     }
 }
diff --git a/...ain/java/de/tudarmstadt/ukp/inception/support/xml/sanitizer/ExternalPolicyCollection.java b/...ain/java/de/tudarmstadt/ukp/inception/support/xml/sanitizer/ExternalPolicyCollection.java
@@ -32,6 +32,7 @@ public class ExternalPolicyCollection
     private boolean debug;
     private ElementAction defaultElementAction;
     private AttributeAction defaultAttributeAction;
+    private String defaultNamespace;
 
     public String getName()
     {
@@ -102,4 +103,14 @@ public AttributeAction getDefaultAttributeAction()
     {
         return defaultAttributeAction;
     }
+
+    public String getDefaultNamespace()
+    {
+        return defaultNamespace;
+    }
+
+    public void setDefaultNamespace(String aDefaultNamespace)
+    {
+        defaultNamespace = aDefaultNamespace;
+    }
 }
diff --git a/...main/java/de/tudarmstadt/ukp/inception/support/xml/sanitizer/PolicyCollectionBuilder.java b/...main/java/de/tudarmstadt/ukp/inception/support/xml/sanitizer/PolicyCollectionBuilder.java
@@ -19,6 +19,7 @@
 
 import static de.tudarmstadt.ukp.inception.support.xml.XmlParserUtils.caseInsensitiveQNameComparator;
 import static de.tudarmstadt.ukp.inception.support.xml.sanitizer.ElementAction.PASS;
+import static org.apache.commons.lang3.StringUtils.isEmpty;
 
 import java.lang.invoke.MethodHandles;
 import java.util.LinkedHashMap;
@@ -46,6 +47,8 @@ public class PolicyCollectionBuilder
     private ElementAction defaultElementAction = ElementAction.DROP;
     private AttributeAction defaultAttributeAction = AttributeAction.DROP;
 
+    private String defaultNamespace;
+
     public static PolicyCollectionBuilder caseSensitive()
     {
         return new PolicyCollectionBuilder(LinkedHashMap::new);
@@ -66,6 +69,12 @@ public PolicyCollectionBuilder(Supplier<? extends Map> aMapSupplier)
         globalAttributePolicies = mapSupplier.get();
     }
 
+    public PolicyCollectionBuilder defaultNamespace(String aDefaultNamespace)
+    {
+        defaultNamespace = aDefaultNamespace;
+        return this;
+    }
+
     public PolicyCollectionBuilder defaultAttributeAction(AttributeAction aDefaultAttributeAction)
     {
         defaultAttributeAction = aDefaultAttributeAction;
@@ -137,6 +146,10 @@ PolicyCollectionBuilder elementPolicy(QName aElement, ElementAction aAction)
         elementPolicyBuilders.put(aElement,
                 new ElementPolicyBuilder(aElement, aAction, mapSupplier));
 
+        if (isEmpty(aElement.getNamespaceURI()) && defaultNamespace != null) {
+            elementPolicy(new QName(defaultNamespace, aElement.getLocalPart()), aAction);
+        }
+
         return this;
     }
 
@@ -201,6 +214,12 @@ void attributePolicy(QName aElementName, QName aAttributeName, AttributePolicy a
                         aElementName, aAttributeName, oldPolicy, aPolicy);
             }
         }
+
+        if (isEmpty(aElementName.getNamespaceURI()) && isEmpty(aAttributeName.getNamespaceURI())
+                && defaultNamespace != null) {
+            attributePolicy(new QName(defaultNamespace, aElementName.getLocalPart()),
+                    new QName(defaultNamespace, aAttributeName.getLocalPart()), aPolicy);
+        }
     }
 
     public void allowAttribute(QName aAttribute, Pattern aPattern)

diff --git a/...main/java/de/tudarmstadt/ukp/inception/support/xml/sanitizer/PolicyCollectionIOUtils.java b/...main/java/de/tudarmstadt/ukp/inception/support/xml/sanitizer/PolicyCollectionIOUtils.java
@@ -53,6 +53,9 @@ public static PolicyCollection loadPolicies(InputStream aIs) throws IOException
             policyCollectionBuilder
                     .defaultAttributeAction(externalCollection.getDefaultAttributeAction());
         }
+        if (externalCollection.getDefaultNamespace() != null) {
+            policyCollectionBuilder.defaultNamespace(externalCollection.getDefaultNamespace());
+        }
 
         for (ExternalPolicy policy : externalCollection.getPolicies()) {
             var isElementPolicy = policy.getElements() != null;

diff --git a/...ain/java/de/tudarmstadt/ukp/inception/support/xml/sanitizer/SanitizingContentHandler.java b/...ain/java/de/tudarmstadt/ukp/inception/support/xml/sanitizer/SanitizingContentHandler.java
@@ -52,6 +52,7 @@ public class SanitizingContentHandler
     private final Logger log = LoggerFactory.getLogger(getClass());
 
     private static final String MASKED = "MASKED-";
+    private static final String NO_LOCAL_NAME = "NO-LOCAL-NAME";
     private static final String XMLNS = "xmlns";
     private static final String XMLNS_PREFIX = "xmlns:";
 
@@ -119,15 +120,20 @@ private void startElement(QName aElement, Attributes aAtts, Optional<ElementPoli
         throws SAXException
     {
         QName element = aElement;
+        ElementAction action = aAction;
 
-        if ((aAction == ElementAction.DROP || aAction == ElementAction.SKIP)
-                && policies.isDebug()) {
+        if (StringUtils.isBlank(element.getLocalPart())) {
+            action = ElementAction.SKIP;
+            element = new QName(element.getNamespaceURI(), NO_LOCAL_NAME, "");
+        }
+
+        if ((action == ElementAction.DROP || action == ElementAction.SKIP) && policies.isDebug()) {
             element = maskElement(element);
         }
 
-        stack.push(new Frame(element, aPolicy, aAction, aLocalNamespaces));
+        stack.push(new Frame(element, aPolicy, action, aLocalNamespaces));
 
-        if (aAction == ElementAction.PASS || policies.isDebug()) {
+        if (action == ElementAction.PASS || policies.isDebug()) {
             super.startElement(element.getNamespaceURI(), element.getLocalPart(), getQName(element),
                     aAtts);
         }
@@ -244,10 +250,15 @@ private void sanitizeAttribute(AttributesImpl aSanitizedAttributes, QName aEleme
         var action = policies.forAttribute(aElement, attribute, type, value)
                 .orElse(policies.getDefaultAttributeAction());
 
-        if ("xmlns".equals(attribute.getPrefix())) {
+        if (XMLNS.equals(attribute.getPrefix())) {
             action = AttributeAction.PASS;
         }
 
+        if (StringUtils.isBlank(attribute.getLocalPart())) {
+            action = AttributeAction.DROP;
+            attribute = new QName(attribute.getNamespaceURI(), NO_LOCAL_NAME, "");
+        }
+
         switch (action) {
         case PASS:
             aSanitizedAttributes.addAttribute(uri, localName, qName, type, value);